14:05:04 #startmeeting Test and Performance Weekly Meeting 14:05:04 Meeting started Thu Jun 16 14:05:04 2016 UTC. The chair is mbeierl. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:05:04 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:05:04 The meeting name has been set to 'test_and_performance_weekly_meeting' 14:05:27 #topic Roll Call 14:05:29 #topic call role 14:05:29 #info kubi 14:05:31 #info Mark Beierl (StorPerf) 14:05:38 #info Yujun Zhang 14:05:39 #info Morgan Richomme (Functest) 14:05:39 #chair morgan_orange 14:05:39 Current chairs: mbeierl morgan_orange 14:05:49 any other takers for chair? 14:06:51 #info Julien 14:07:24 #topic QTIP Presentation 14:07:57 I am doing #topic incorrectly too? 14:08:05 #help 14:08:08 #info Maryam Tahhan 14:08:15 on IRC only today 14:08:30 mbeierl: I think you have to be chair 14:08:31 #link https://wiki.opnfv.org/display/qtip 14:08:35 #info QTIP being discussed 14:09:11 mtahhan: I am chair, but it does not seem to be responding to help or topic... Maybe collabot is a little sleepy 14:09:13 mbeierl, You're doing it right, I think 14:09:51 mtahhan: the QTIP link ^ is what is being displayed on the GTM. 14:10:15 #info discussion of QTIP scope 14:11:03 I lost the audio 14:11:14 #info open question after last demo: is QTIP going to continue independently or should it roll up into Functest and Yardstick 14:11:17 I think there was a similar issue with meetbot just last week 14:11:24 dun dun duuuuuuun 14:11:36 mtahhan: I think so... 14:11:37 #info morgan_orange mentioned what's the overlap between qtip and other projects such as bottleneck 14:12:09 #info yujunz discussed QTIP internally and is proposing a roadmap. 14:12:24 Julien-zte: my concern is more with yardstick, bottlenecks is a bit different 14:13:02 #link https://gerrit.opnfv.org/gerrit/#/c/15681/4/docs/roadmap.rst,unified 14:13:03 morgan_orange, qtip care about the performance 14:13:06 #info new PTL, new members, so new roadmap is proposed, please comment on the roadmap 14:13:50 #info morgan_orange asked if QTIP tests *before* the OPNFV installation? 14:13:57 from the begin, yardstick and functest is more about the feature test 14:14:03 Julien-zte: yardsitck also care about performance 14:14:21 I don't know the scope changes 14:14:50 Julien-zte: performance testing is the key testing area since yardstick setup 14:14:57 #info yardstick has now confirmed as being umbrella for all performance testing - the ones built in, and well as the interface to CPerf, StorPerf, and VSPerf 14:15:10 Julien-zte: Functest = functional testing, Yardstick = ërformance 14:15:21 #info Functest is umbrella for all functional tests 14:15:47 I can't hear anythiing 14:16:00 from the begin, functest test on the infrastructure, yardstick perform the test as a tenant in a VM 14:16:28 it is the place we can discuss more clear about the overlap 14:16:48 let's avoid overlapping work in the future 14:18:02 #info During Colorado planning, there has been a discussion of test projects to attempt to clear the overlap. It comes down to all projects should be affiliated with one of the two umbrellas: performance (Yardstick) or functional (Functest) 14:18:57 who can provide the info about yardstick become the umbrella for other test projects 14:19:04 #info that does not mean that individual projects go away. VSPerf is very much independent, but it's OPNFV reporting and CI pipeline are going to be driven from Yardstick integration 14:19:32 mtahhan: can clarify that statement if needed. I just don't always want to talk about myself 14:19:34 hopefully :P 14:19:43 ditto for StorPerf 14:19:55 Julien-zte: let talk in berlin 14:20:00 good 14:20:02 Sure 14:20:39 #info agreed that this conversation will continue in Berlin: "How does QTIP fit in now that there is the plan to have the two major umbrella projects" 14:21:20 mbeierl, morgan_orange, let's have a discuss on the summit 14:21:29 and kubi001 14:21:59 #link https://wiki.opnfv.org/display/functest/testing 14:22:21 Discussing the overview picture 14:23:31 #link https://wiki.opnfv.org/download/attachments/2926690/Testing.png?version=1&modificationDate=1463478046000&api=v2 14:24:20 #link https://wiki.opnfv.org/download/attachments/2926690/how_to_testing.png?version=1&modificationDate=1464962177000&api=v2 "How to interact with other projects" 14:24:32 #topic "Conversations with the test teams" 14:25:01 #info to introduce ourselves to attendees - perhaps upstream contributors 14:25:24 #info to discuss our goals and ambitions (collective and individual) for the testing projects 14:25:48 Hey folks... I would suggest that we re-draw this picture to make it comprehensive - i.e. include the earlier presented Qtip. 14:26:03 In addition, does Bottlenecks really feed into Yardstick? 14:26:31 frankbrockners: During the plugfest, it appeared that Bottlenecks can use information from Yardstick 14:26:45 so it draws info, not publishes info 14:31:16 #info discussion about the https://wiki.opnfv.org/download/attachments/2926690/Testing.png?version=1&modificationDate=1463478046000&api=v2 picture 14:32:02 #info it's a way of looking at the testing framework 14:32:24 #info Presentation is in git, team leads to add content 14:32:50 #info Goal is to emphasize Functest is upstream focused 14:33:45 #info, and that it's time to give back :) 14:34:04 #info Open the discussion about defining Telco KPIs 14:35:03 mbeierl - thanks - so sounds like the arrow points the wrong direction 14:36:07 mbeierl - I also thought that vsperf publishes directly into MongoDB - and does not go via Yardstick to do so - but mtahhan might correct me here 14:36:35 frankbrockners: I think so. Either that or Bottlenecks indicated it's intention to read from different sources and use Yardstick for reporting to the common DB. 14:37:19 frankbrockners: you'll notice that VS, C, and Stor-Perf all have double arrows. Yardstick can kick them off, and they can report back up 14:37:35 mbeierl, in the summit we can discuss about the where to store and how to demonstrate the result 14:38:41 #info Daniel clarified that the diagrams do not constitute formal organization, but rather a mutual agreement on presentation of how things fit together 14:39:53 mbeierl - thanks - also just checked with mtahhan that vsperf will continue to publish directly into mongodb - so the real picture would look a bit more "complex" :-) 14:40:39 frankbrockners: yardstick can publish to mongodb/influxDB 14:41:07 but that is why we put several DBs in the figure 14:41:36 yup -- but you probably need more arrows in your picture ;-) 14:41:37 right, but to frankbrockners point, there are no "direct lines" from the *Perf projects to the DB 14:41:49 ah ok right 14:41:54 frankbrockners: vsperf can publish directly into mongodb, it also can go via Yardstick when Vsperf integrate with Yardstick. It depends on Vsperf 14:42:41 same with Storperf 14:42:45 kubi001: that is also a question that StorPerf has - we use a private Carbon (Graphite) DB 14:42:56 kubi001: exactly, thanks! 14:44:37 mbeierl: Carbon DB + Grafana in StorPerf? 14:45:15 kubi001: correct: https://wiki.opnfv.org/display/storperf/Storperf+Architecture 14:46:32 mbeierl: great 14:46:41 kubi001: Right now, I summarize the findings into an average at the end of the run, and only publish that summary result to the MongoDB via the test API 14:47:42 kubi001: but I need to decide where the summary should be done - in StorPerf, or in the Dashboards later? How much data do I want people to store in to common DB? This can be gigabytes worth of data per run... 14:48:45 My thought is: StorPerf exports the Graphite (and I one day will add Grafana) via a web UI. The raw data can be examined in the container. However, the final test result will be the summary only to save space. 14:50:36 mbeierl: I see. Same situation with yardstick, so we support MongoDB via test API, and also support InfluxDB + Grafana to support more detail result 14:51:39 dfarrell07, mtahhan: what are you planning for your raw data vs. summary results for longer term storage and publication? 14:54:01 mbeierl: we haven't gotten that far, but we don't expect to have tons of data for the tools we use so far, so we will not have that problem 14:54:12 nice :) 14:55:00 :) 14:55:41 #action: morgan_orange to create a page for Berlin summit test group planning 14:59:15 #info Will discuss during the summit how to have rolling organization of this weekly meeting so that it becomes more collaborative 14:59:48 #info Project leads, etc, encouraged to fill in details in the presentation 15:00:02 bye 15:00:10 #endmeeting